Stage III annealing kinetics have been studied for pure and radiation-doped copper, the latter containing a high concentration of interstitial and vacancy clusters which act as efficient sinks for mobile vacancies in stage III. In the doped samples, the influence of secondary defects such as divacancies on the annealing kinetics is suppressed, and the measured activation energy, 0.70 ± 0.03, must be interpreted as the migration energy of single vacancies. In undoped samples the same activation energy is found although the formation of vacancy clusters in stage III indicates that highly mobile and tightly bound di- or multiple vacancies are also present. This is consistent since the present method of activation energy determination cannot reveal an influence of highly mobile divacancies. Such an influence can be important to explain the observed second order dose shift of stage III for undoped samples.